Home › Stats › Hypothesis tests

TTestResult

A t-test's result: the statistic, the p-value and the degrees of freedom.

public sealed record TTestResult(double Statistic, double PValue, double DegreesOfFreedom)

Properties — Statistic is the t statistic. PValue is the p-value on the requested tail. DegreesOfFreedom is the degrees of freedom: integral for Student and for the paired and one-sample tests, fractional for Welch, whose Satterthwaite denominator is not a count of anything. It is one for Welch on two constant samples, whose denominator is 0/0, and NaN when NanPolicy.Propagate kept a NaN, as scipy's are; ConfidenceInterval is then NaN at both ends.

Example — a one-sample test's degrees of freedom: one less than the number of values actually tested, sample.Length - 1 here since the default NanPolicy.Propagate tests every value — one less than the filtered length under NanPolicy.Omit instead.

using Lodestar.Stats;

double[] sample = [12.1, 9.4, 15.0, 11.2, 8.8, 13.9, 10.5];

TTestResult result = TTest.OneSample(sample, populationMean: 10.0);

double df = result.DegreesOfFreedom;   // => 6

Remarks — DegreesOfFreedom is a double, not an int, because TTest.Independent under Variance.Welch needs to report a fractional value; every other entry point happens to land on a whole number, which is why this example's 6 prints with no decimal point at all — it is still the same double field.

Alongside the three public properties, a TTestResult privately carries the estimate its test compared and its standard error — internal, since scikit-learn-style callers only reach them through ConfidenceInterval, which is where they surface.

Being a record, two results with the same statistic, p-value and degrees of freedom are equal.

Applies to — net10.0, netstandard2.0.

See also — TTest.Independent, TTest.Paired, TTest.OneSample, TestResult, the Python equivalence table.

Members

Member	What it does
TTestResult.ConfidenceInterval	The confidence interval for the difference this test measured.